XPS STUDY OF GAS SENSITIVESn02 THIN FILMS V.Brinzari*, G.Korotcenkov*, K.Veltruska, V.Matolin, N.Tsud, J.Schwank** :>
*
Dep. of Chemical Engineering, University of Michigan, Ann Arbor, USA Dep. of Electronics and Vacuum Physics, Charles University, Prague, Czech Republic * Department of Microelectronics, Technical University of Moldova, Bld. Stefan cel Mare, 168, Chisinau, 2004, Moldova. E-mail:
[email protected]
ABSTRACT Sn02 is a dominant material in production of solid state gas sensors. However, till now the nature and forms of chemisorbed species on its su$ace remain debatable. In this report we present the results of XPS study of the Sn02 thin films after various treatments in vacuum, oxygen, and both dry and humid atmospheres (T,,2=25-600"C}.SnOz films were deposited by spray pyrolysis method. We analyzed the changes of th e Sn02 su$ace stoichiometry, energy positions of main X P S peaks, and concentration of adsorbed. particles. On the base of obtained results the assumptions on the nature of absorbed species on the Sn02 su$ace have been made.
1. INTRODUCTION X-ray photoelectron spectroscopy (XPS) is the effective and sensitive method for of surface physical-chemical interaction study, and it successfully applied for investigation of metal oxides. XPS studies of tin dioxide showed, that the complicated structure of 01s oxygen line is connected with different chemical forms of oxygen, coexisting on the semiconductor surface. Generally, two additional convoluted peaks are observed at high-energy part of 01s binding energies (BE). These peaks are marked usually as I-st, 11-nd, and 111-rd peaks in order of increasing binding energies. The I-st one ( O(1) ) corresponds to the basic peak of Sn02 lattice However, ~). the nature of the 11-nd oxygen ( 0 1 ~ and the 111-rd peaks (O(I1) and O(II1) ) remains debatable, so far. The widely used explanation of peaks, mentioned above, is based on the presence of two forms of water, adsorbed on the surface [I]. 11-nd peak is considered to be of OH groups origin, and the 111-rd one corresponds to H 2 0 molecules. Other authors [2,3] link up these
peaks to chemisorbed oxygen in Or!or 0 forms, and to physisorbed water or physisorbed oxygen.
2. EXPERIMENT SnO? thin polycrystalline films with thickness 30-70 nni were deposited by spray pyrolysis method, using 0,2M SnCI4-water solution (T,,, = 35O-50OoC)[4].For Sn02 surface doping by Pd we applied the same method of spray pyrolysis from water solution of PdC12 (Tpyr= 430OC). For Sn02 deposition we used low resistivety Si substrates to minimize charging effect during XPS experiments. XP spectra were collected on the ultrahigh vacuum system equipped with hemispherical electrostatic analyzer Omicron EA 125. In this work we used MgKa line (hv =1253.6eV) for excitation. The Ols, Sn3dSl2and C l s lines were recorded in detail with the pass energy of 20 eV. The standard fitting procedure was used to analyses spectra and to separate three overlapping peaks of 01s line. We used several treatment procedures during sample examinations. At first, SnO? films were placed in vacuum chamber and annealed at temperatures 100, 160, 200, 420, and 600°C during 15 min. Then, after XPS measurements, these samples were treated in dry and humid air at T=300"C during 10 min. After treatment in humid air the samples were pre-annealed in vacuum conditions (T=l20"C, t=15 min) before' recording spectra. Such annealing excluded the presence of water in molecular form on the Sn02 surface. Ar' sputtering of Sn02 films (E=3 keV; P=10-7Pa), and annealing in oxygen atmosphere (30% 02+He) were used as additional treatments of analyzed SnO2 samples. XPS spectra were recorded after each treatment. The relative atomic surface concentrations (ASC) of chemical species (O;,ds; Olat;Sn; Pd and C) were calculated by using both
0-7803-5885-6/00/$10.00 0 2000 IEEE
127
than in humid one. Annealing i n oxygen atmosphere makes this influence morc clcar. Such treatment enhances the intensity 0 1 Il-ncl peak and suppresses the intensity of 111-rd pcali (see Fig.3).
peak areas and standard atomic sensitivity factors
PI. 3. RESULTS The main results, which we obtained from our experiments, were the following: 1 . All recorded 0 1 s lines exhibit three quite well resolved peaks (Fig.1, peaks I, 11, and 111). We confirmed the "surface" nature of 01s additional peaks (I1 and 111), positioned in the range of higher BE. XPS spectra, measured at two different angles between the sample surface and axis of analyzer input, illustrate this fact in Fig.1.
100 300 500
A
Treatments. C
B
Fig.2. Inllucnce of various thermal treatmcnls on ~ l i c parameters of' SnOl X P spectra (cp = 90"): A- d r y ikii.: T,,,=300"C; L=IO min; ]U- humid air; T~,,,=300"C:I = IO min).
533 531 529 527 Binding energy, eV Fig.1. Typical 0 1 s XP spcctrum of SnOz thin film: 1 cp = 90°; 2 - cp = so.
The effective photoelectron output depths i n these two cases are equaled 10 (cp =90") and 3 (cp =8") atomic layers approximately. Thus, the contribution of adsorbed species in total intensity of 0 1 s peak increases considerably in second case. The obtained results show that ASC of species responding to 01s (11-nd, 111-rd peaks) and Cls lines have rather large values, which are much higher than 0, I ML. 2. The interesting feature can be seen on temperature dependencies of 11-nd 01s peak's normalized intensity O(II)/O(I) (see Fig.2). The sharp decreasing of intensity takes place up to 200'C. At higher temperatures the 11-nd peak intensity dropping is rather smaller. The 111-rd 01s peak doesn't have such a behavior. The traces of species, corresponding to these peaks, are observed even till 600°C. The intensity of IInd peak is more affected by annealing in dry air,
A
B
C
D
Fig.3. Correlation bctween normalized iiitcnsitics 01' O(I1) and O(II1) peaks of 0 1 s oxygen linc and various SnOz surface treatments (cp=S"j: A- i n i t i a l state; U- Arf sputtering, (=3 min; C- exposition iii air at RT, [=Mays; D- sputtering, t=l min.; F- anncnling in oxygen atmosphere, T,,,,=250"C, t=30 min
It's necessary to note that thc m i i i n change i n BE of Sn3d, 0 1 s and CIS lines chi-ing annealing in the vacuum takes place at thc S;II.IIC temperature range RT - 200°C (Fig.?). Annealing of SnO? films i n vacuum shil'ts the 13E of all lines in the range of' lower energics. The average shift of BE is equaled 0?3-0.35 eV. The same shift of the base peaks in SnOz XP spectra is observed also after Ar'. sputtering of SnO. surface without any thermal treatment. 11 shows
128
that the BE shift iii both these cases has the same nature. 3. During the annealing in vacuum conditions the intensities of lattice 01s and Sn3d peaks increase simultaneously with decreasing of Cls peak intensity (Fig.4). Annealing at usual conditions (dry and humid air) easy restore the initial state of SnO? surface. The numerical estimations of atomic surface concentration (ASC) of carbon species on the SnO, surface give the values, which are more than 0,S-0,7 of monolayer. Energetic position of Cls peak (BE-284,s eV) corresponds to graphite phase of carbon. Carbon C l s line practically has not additional peaks, attributed to other carbon containing species. Only small traces close to background at BE-290 eV were revealed.
peak has energetic position 336,l eV, which corresponds to PdO phase. It shows that at u\ed TPY,=430"C the initial and gaseous produc~sof the reaction of PdCI? pyrolysis are not observed on the SnO? surface. a
$
a,
500 -
n
U
'2 400-
-
CO
in-
% 300 a 5 0
-$200
"'
...f
' io
i,
do go 60 0 2 surface concentrationof carbon, arb. r
Fig.5. Dependencies of intcnsities of XI'S lntticc Sn3d and 01s peaks on surface carbon conccntratioii. I i n '
100
300 500
A
B
Treatments, C
Fig.4. Inlluence of various thcrinal treatments on the intcnsities of niain peaks of SnOz XP spectra (cp=OO"): A- dry air; T,,,,=300"C; t = I O niin; U- humid air; T,,,,=300"C;(=I 0 inin).
4. Correlation between carbon surface concentration and Sn3d and 01s peak intensities, responding to SnO? lattice atoms, showed that intensity of Sn3d peak has more strong Sn02 surface carbon species concentration dependence (see Fig.5). Such different influence of treatments (annealing, sputtering) on the intensities of 01s and Sn3d peaks, corresponding to lattice 0 and Sn in the SnO,, requires careful using of Sn/O ratio for estimation of correct SnO? stoichiometry. Besides that it allows to assume that carbon 011 the SnOz surface is bonded predominantly to Sn. 5. In XPS spectra of SnO2:Pd the peaks connected with presence of PdC12 and metallic Pd on SnOl surface are not observed. XPS Pd3d
6. Surface doping of SnO? filins by Pd changes the shape of 0 1 s line. 0 1 s spectra for Sn02:Pd samples have only one additional peak, positioned at the same BE as 11-nd 01s peal; f'or undoped Sn02 films. Besides that, XPS spectra for doped films (positions of Ols, Sn3d, Cls peaks) are shifted on 0,2-03 eV in the range of lower BE in comparison with similar spectra for undoped SnO? films.
4. DISCUSSIONS We can affirm that i n our experiniental conditions the nature of 111-rd 0 1 s peak is not connected with presence of adsorbed w:i!er (H20), because this peak was observed at temperatures, when water couldn't exist on the surface due to desorption processes. Bond energy for water is equaled - 80-100 kJ/mol. Nonmonotonous temperature behavior of' 11-ncl peak (see Fig.2) points out on its iiiore complicated structure. We assume that two oxygen containing species with different hond energies with SnO? surface are responsible lor this peak. Change of BE with increasing temperature (Fig.2) points out on decrense of work function and, evidently, on surface Ixind bending. This effect occurs mainly up to 200°C. Species, responsible for such mechanism. should be acceptor-like ones. Neither H1O molecules, nor OH-groups exhibit such properties i n given
129
temperature range. Obviously they are weak donor-like species. The character of behavior of O(1) and O(I1) peak intensities with annealing in pure oxygen and wet atmosphere (Fig.2, 3) allows to suppose that part of species, corresponding to IInd peak, is molecular oxygen ( 0 2 ) , because the transition between molecular and atomic oxygen takes place on metal oxide surface at the temperatures 170-200°C. So, other part of species responsible for O(I1) peak, especially at T>200°C, can be in the form of atomic oxygen or OH-groups. The last one can exist at least in two forms, such as "rooted" OH-groups, bounded with Sn atoms, and OH-groups, which can be formed during reaction of H20 dissociation with participation of so-called "bridging" lattice oxygen H , O ~+SnOz c1 (sn4+ -OH-)- +(o& - H + > TPD experiments [6] confirm the existence of various forms of OH-groups under hydroxylation of SnOpsurface. It seems that the nature of another 01s peak (11-rd peak) can be also connected with OHgroups, which can have various forms with different bond energies. We suppose that OHgroups, forming the 111-rd peak, are "bridging" origin However, OH-groups, forming 11-nd peak, are the "rooted" OH-groups (besides of contribution of chemisorbed oxygen). Such model allows to explain the shift of BE under Pd doping. In given case we have lower amount of hydroxylation (absence of 11-rd peak). It leads to the electron affinity decrease, due to less dipole component and, therefore, to decrease BE for Pd doped SnO:!. It is rather difficult to explain the behavior of carbon species. From one hand, the shape of C l s line indicates only on atomic nature of carbon on the Sn02 surface. We don't observe on XPS spectra any carbon groups like carbonates and other one. From the other hand, the initial concentration of carbon (before sputtering and annealing in vacuum) is easily recovered after annealing at ambient conditions (T-300°C). One can suppose that carbon dioxide, present in the air, is the only source for such species. Therefore, for explanation of appearance of atomic carbon (graphite phase) on the Sn02 surface, we have to suppose that the dissociation of CO:! molecules takes place on the Sn02 surface even at RT. This effect has not been
reported earlier. The position of carbon atom is over the lattice tin atom (Fig.5 confirms this fact).
5. CONCLUSIONS We demonstrated that the nature of adsorbed species on the SnO:! surface in basic state of gas sensitive film, i.e. without active gas, is rather complicated, than it has been reported earlier. We think that several oxygen containing adsorbed species ( 0 2 , 0, OH, H20) can be responsible for the same peak of 01s peak position in XP spectra of Sn02. At that, the contribution of these individual species in total intensity of O(I1) or O(II1) peaks depends on the state of SnO:! surface. Such situation makes more difficult the interpretation of the XPS results, obtained by XPS, and makes it ambiguous. Therefore, for better understanding of the SnOz suriace chemistry it is necessary to use combined approach, which would include, besides XPS, such methods as TPD, IR- and Raman spectroscopes, EELS and so on.
ACKNOWLEDGMENTS The part of this work was supported by the EC in the frame of "INCO-COPERNICUS" Program.
REFERENCES [ 11. G.Gaggiotti, A.Galdikas, S.Kaciulis, G.Mattogno, A.Setkus. Sensors and ActLrators. B.24-25 ( I 995) 5 16-5 19. [2]. P.Serrini, V.Briois, M.C.HorriIlo, A.Traverse, L.Manes. Thin Solid Films. 304 (1997) 113-122. [3]. W.K.Man, H.Yan, S.P.Wong, T.K.S.Wong, 1.H.Wilson. Proc. of MRS Symposiuin, Boston, USA, 27 Nov.-1 Dec. 1995. V.403, 441-446. [4] G.Korotcenkov, V.Brynzari, S.Dmitnev, J. Muter. Sci. And Engineering. B.63(3) ( 1999) 195-204. [ 5 ] . C.D.Wagner, W.M.Riggs, L.E.Davis, J.F.Moulder, G.E.Muilenburg, Hanhook of XRay Photoelectron Spectroscopy (Perkin-Elmet-, Eden Prairie, MN, 1979). [6] M.Egashira, M.Nakashima, S.Kawasumi, J.Phys. Chem. 85 (1981) 4125-4130.
130
